Ethnicity and acute hospital admissions: Multi-center analysis of routine hospital data

Abstract

Background: The effects of ethnic and social inequalities on patient outcomes in acute healthcare remain poorly understood. Methods: Prospectively-defined analysis of registry data from four acute NHS hospitals in east London including all patients ≥ 18 years with a first emergency admission between 1st January 2013 and 31st December 2018. We calculated adjusted one-year mortality risk using logistic regression. Results are presented as n (%), median (IQR), and odds ratios (OR) with 95% confidence intervals. Findings: We included 203,182 patients. 43,101 (21%) patients described themselves as Asian, 21,388 (10.5%) Black, 2,982 (1.4%) Mixed, 13,946 (6.8%) Other ethnicity, and 100,065 (49%) White. We excluded 21,700 (10.7%) patients with undisclosed ethnicity. 16,054 (7.9%) patients died within one year. Non-white patients were younger (Asian: 43 [31–62] years; Black: 48 [33–63] years; Mixed 36 [26–52] years) than White patients (55 [35–75] years), with a higher incidence of comorbid disease. In each age-group, non-white patients were more likely to be admitted to hospital. This effect was greatest in the ≥ 80 years age-group (32% non-white admitted to hospital versus 23% non-white in community population). Deprivation was associated with increased mortality in all ethnic groups (OR 1.41 [1.33–1.50]; p < 0.001). However, when adjusted for age, Asian (0.69 [0.66–0.73], p < 0.0001) and Black patients (0.79 [0.74–0.85]; p < 0.0001) experienced a lower mortality risk than White patients. Interpretation: Ethnic and social disparities are associated with important differences in acute health outcomes. However, these differences are masked by statistical adjustment because patients from ethnic minorities present at a younger age. Funding: None

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 23 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.